What reference ranges mean

Most blood tests are performed with commercial kits. The manufacturer of each kit is responsible for advising the appropriate ranges of results.

Typically they will have performed the test on a number of supposedly "healthy" people to determine the acceptable range of results. The people concerned may have been admitted to hospital for elective (non urgent) surgery, or worked for the company concerned or been blood donors. They hopefully would have been in the FAQ category re. their health.

They probably didn't belong to any particular ethnic group, and may not have included people at extremes of age, pregnant etc.

Many variables may be such that the reference range quoted is not even slightly relevant to you in particular.

These variables are in part responsible for some tests having very broad ranges quoted.

An example is vitamin B12 levels. My labs have different kits, so slightly different ranges. Both are very wide ranges, in SI units from 150 to 700. What is healthy? I'm pretty happy if the result is over 400, but don't rely on this alone even then. I always have a homocysteine measurement as well, to double check.

Homocysteine is quoted at about 3 to 14 (SI) but I'm unhappy if it is over 8. There are a lot of people walking around with higher homocysteine levels due to subclinical vitamin B12 deficiency. They have obviously been included in the reference range determination.

You are your best "control." Compare tests with previous results.

It will save any future doctor a lot of time and trouble if you are able to provide this information anytime the same test is repeated.

Sometimes the gradual change in a test over time, is a vital clue. An example of this is the MCV. Your red blood cells live for about 100 days, and become progressively smaller as iron deficiency developes, larger as vitamin B12 or folic acid become deficient.

This can happen without the values leaving the quoted range.

Then along came Sam Queen

Statistical stuff

On any one occasion these variations may be at their peak and in the same direction, so as to, say, reduce the result of the test.

If the same test is repeated on another day, the chance that the variables will be maximal and in the same direction, is pretty unlikely.

The result will therefore be closer to average. This is called "regression to the mean." Any extreme result is likely to be closer to average the next time it is repeated.

The ranges quoted, assuming equal spread above and below the average, usually¹ include 95% of the reference population². If you have a lot of tests the chance that one of them will be outside this range, increases. If you had 20 tests, that chance would be 64%.

Need help or have a helpful tip to share, or an experience others can learn from?

You can have your own page on this site, be anonymous if you prefer, and help other people or get useful opinions.

From reference ranges page back to the main index page

Notes and references

2. This is the average +/- two standard deviations, in a Gaussian distribution (the classical bell-shaped curve, with most items near the average, progresively fewer the further away from it in either direction.)